learning, adapting, modifying, and perfecting previous understandings.
However, information that is inconsistent with initial causal attributions may trigger a reinterpretation of causal attributions. We suggest that double-looped learning-oriented information processing systems encourage reinterpretation (Meyer, 1982).
Third, the interpretations must lead to some form of action (Meyer, 1982), which may or may not be consistent with previous behavior. If new information has not caused much reinterpretation, the organization may not change much beyond fine-tuning its existing action course. Conversely, with reinterpretation, the organization has an opportunity to test the new causal attributions with a changed course of action. Information processing systems that allow for new action, then, are more likely to stimulate fresh learning (Lant et al., 1992). The results of the action provide further feedback information to the organization that can potentially be noticed as the information processing cycle continues. It becomes clear why learning systems are couched in terms of a ‘‘loop!’’
Further, learning can be enhanced if information processing speed is increased, thereby allowing more iterations through the noticing, reinter- pretation, new action, and feedback cycle. As the cycle progresses, information is converted into knowledge and these knowledge assets are added to the firm’s storehouse of learning. The above suggests that information processing systems that facilitate broader and deeper noticing, reinterpretation, new action, and multiple iterations promote radical, double loop learning and cause information processing to be a value- adding activity that can correct initial misperceptions.
Organizational information processing cycles may be dominated by a tendency to reinforce existing interpretations, i.e., they often generate only single-looped learning. It is the more unusual system that can encourage double-looped learning. Why is the default information processing cycle prone to reinforcing, rather than frame-breaking, outcomes? Again, it makes sense to start looking at individual information processing patterns for insight here. First, individuals rarely notice all the information needed for a decision and they tend to ignore indicators that suggest that their initial action, or interpretation, was ill advised (Hogarth, 1987). Second, if relevant but discordant information is noticed, it may not be interpreted correctly. Input information is often ambiguous and the original conclusions formed from biases tend to persist (Cooper, Woo, & Dunkelberg, 1988).
Third, the firm must have the freedom to adjust its actions in ways that are consistent with the new interpretation if the venture is to succeed. Too much redirected action signals to key stakeholders a confusion of purpose and is
often discouraged in established firms. Further, in established firms, organizational incentives and other reporting systems embed inertial tendencies and resistance to change. Even in new ventures, a headstrong entrepreneur, who has launched a firm in a given direction, may be difficult to redirect (Mone, McKinley, & Barker, 1998). Finally, time may be an issue. Correcting misperceptions may require cycling through multiple iterations of noticing, interpreting, and acting, yet new ventures often have only small temporal windows for action. And unlike the ability to make rapid decisions when heuristics are invoked for the initial opportunity, a changed course of action requires a more time-consuming, complex process of reframing the situation; this new conceptualization explicitly denies the intuition that the heuristic initially provided, and thus, heuristics cannot be relied on at this point to speed up the process.
Using Strategic Factors to Enhance Double Loop Learning Information Processing Systems
At the most general level, then, the above arguments suggest that organizations typically find it difficult to (1) increase the breadth and depth of information noticed, (2) reinterpret data, (3) generate new action, and/or (4) decrease cycle time, because they have limited capacity or interest (Cope, 2005; Miner & Mezias, 1996) to process novel information (Daft & Weick, 1984). Several strategic management scholars, however, assert that key strategic factors may increase the capacity of organizations to process novel information or may decrease the demand placed on the organization’s information processing system (Lant et al., 1992; Meyer, 1982; Thomas &
McDaniel, 1990). It is important, therefore, to identify these factors if one wishes to enhance the ability of an information processing system to generate double loop learning (Meyer, 1982; Milliken, 1990). Specifically, consistent with the literature on new venture performance and strategic management (e.g., Sandberg & Hofer, 1987; Thomas & McDaniel, 1990), we propose that two factors – TMT characteristics and organizational structure – may help the information processing system accommodate double loop learning, which, in turn, moderates the relationship between initial misperceptions regarding an opportunity and venture performance. Although these two strategic arenas, and the specific factors within each, are not exhaustive, we believe they represent a critical set of possible influences on a firm’s information processing system that can enhance performance by modifying the initial perceptions.
The overall model of these relationships is pictured in Fig. 1.
MARK SIMON ET AL.
142
Top Management Team Characteristics
Researchers have investigated TMT characteristics with respect to new ventures and firm performance (e.g., Duchesneau & Gartner 1990; Keeley &
Roure, 1990; Wiersema & Bantel, 1992). We propose that the presence of two TMT factors – TMT experience and TMT functional heterogeneity – may strengthen the double loop learning capabilities of the information processing systems of new ventures.
TMT Experience
Prior research indicates that TMT members who have experience in an industry or experience with new venture formation are more likely to be associated with new venture success (Duchesneau & Gartner 1990). We propose that, even for ventures begun on the basis of misperceptions, TMT experience may also improve new venture performance by fostering a learning-oriented information processing system within the new firm.
The new venture’s TMT may range from being highly experienced to extremely inexperienced in new venture formation. The TMT may also vary in its experience in the new venture’s product market. Strategy researchers have argued that extensive experience in an industry reinforces the manager’s commitment to an organization’s status quo (Hambrick, Geletkanycz, &
Frederickson, 1993) and the organization’s conformity to the central tendency Discontinuous
Change
Venture Performance Inaccuracy of
Initial Perceptions
Organization Structure
Innovative Opportunity
Identified
Top Management Team
Re-Interpretation Width & Depth
of Noticing
Organicity Decentral-
ization Heterogeneity
Experience
Double Loop Learning Information Processing System
New Action
Feedback
P2+ P3+ P4+ P5+
P1-
Fig. 1. The Moderating Role of TMT Characteristics and Organizational Structure on Information Processing Systems in New Venture Formation and Performance.
of the industry (Finkelstein & Hambrick, 1990). We contend that prolonged industry tenure, however, does not necessarily decrease organizational innovation (e.g., Bantel & Jackson, 1989). Further, Hambrick et al. (1993) acknowledge that teams with established industry wisdom can benefit those who have new ideas because that wisdom can inform the new ideas. In new venture formation, we believe experience may enrich the learning orientation of the TMTs. For instance, experienced TMTs may have access to more information sources, such as key customers, suppliers, and even competitors.
From a cognitive perspective, experienced team members may be more able to notice subtle differences in information patterns that naı¨ve observers may overlook (Fiske & Taylor, 1991; Gaglio & Katz, 2001) and notice information more quickly than others because of their easier access to information sources and their ability to rapidly assimilate incoming information. Thus, TMTs with new venture formation or product/market experience may have broader, faster, noticing capabilities than other teams.
Experienced TMTs may also be able to reinterpret information more readily than other teams. Experience has given managers a complex system of causal relationships that have already been through substantial revision.
Therefore, experienced TMT members can avoid the novices’ false starts regarding causal relationships. Reinterpretation for experienced managers is faster and can be at a deeper, more refined level of causal attributions than for inexperienced managers, leading to accurate perceptions more quickly.
Eisenhardt (1989), for instance, found that TMT experience is important when one is likely to encounter inaccurate information, suggesting that executives with extensive industry experience can provide high-quality feedback. TMTs with experience in either product/market or venture formation, therefore, are better able to correct misperceptions than teams that lack experience in these areas.
Finally, experienced TMTs may have more time to cycle through the information processing system than other teams because they may be perceived as more legitimate. This legitimacy may give the team more credibility, so, when their misperceptions are apparent they may get more leniency from their stakeholders, in terms of time and financial resources, to learn and adjust their perceptions of the situation (Stinchcombe, 1965).
And, of course, a team has more person hours to invest per day than the individual entrepreneur, which opens up the possibility that the team could process information more quickly than an individual if it could process parts of the information in parallel.
Consistent with these ideas, high levels of experience may confer ‘‘expert’’
status on certain TMT members. Experts have valuable information MARK SIMON ET AL.
144
processing characteristics that can guide and support a team to use feedback information to form more sophisticated causal understanding of the new domain. For example, Chi (2006) defines an expert as someone who can, with minimal effort, recognize fundamental patterns and their subtle distinctions in order to efficiently interpret new information in an accurate light. According to Chi (2006, p. 22) the expert is ‘‘the distinguished or brilliant journeyman, highly regarded by peers, whose judgments are uncommonly accurate and reliable, whose performance shows consummate skill and economy of effort, and who can deal effectively with certain types of rare and ‘tough’ cases.’’
Thus, TMT experience may lead to a double loop learning information processing system by broadening noticing, allowing for more reinterpreta- tion, processing more quickly, and getting more iterations through the system than other teams. This double loop learning process will be especially beneficial to new ventures when entrepreneurs initially misperceived opportunities, which leads to Proposition 2:
P2. TMT experience will moderate the relationship between inaccuracy of initial perceptions and venture performance: TMT experience will have a more positive effect on the venture performance of entrepreneurs who had inaccurate initial perceptions than on the venture performance of entrepreneurs who had more accurate initial perceptions.
TMT Functional Heterogeneity
Functional heterogeneity of the TMT includes the distribution of functional backgrounds, educational curricula, and/or industry experience (Wiersema &
Bantel, 1992). Functional background heterogeneity has been hypothesized to improve the learning component of team processing because there is an expectation that TMT members with different backgrounds will bring different perspectives, broadening both the noticing and the interpreting components of team information processing (Hambrick & Mason, 1984; Lant et al., 1992).
TMTs with heterogeneous functional backgrounds have been associated with adaptability (Murray, 1989), and better performance (Norburn & Birley, 1988).
Education heterogeneity has also been associated with change (Wiersema &
Bantel, 1992). In the new ventures literature, occupationally heterogeneous teams have been found to outperform other teams (e.g., Keeley & Roure, 1990) and teams with industry experience heterogeneity were associated with new venture growth (Eisenhardt & Schoonhoven, 1990).
We propose that TMT heterogeneity will be particularly important for those entrepreneurs that rely on heuristics based decision making
because having members from diverse backgrounds will increase the number and variety of the approaches/models and theories, thereby increasing the probability of deriving an optimal solution. Further, it may be through its effect on the capacity to experience double loop learning that TMT heterogeneity improves an organization’s ability to change or innovate.
TMT functional heterogeneity may promote a double loop learning information processing system that can correct early errors in perception (Lant et al., 1992). Specifically, these teams may have broader noticing capabilities because team members can draw on a wider variety of expertise and information bases, which leads to a diverse set of views (Bantel &
Jackson, 1989; Hambrick & Mason, 1984). In addition, functionally heterogeneous TMTs may be more likely to have reinterpretations of their initial causal attributions because their differing perspectives provide natural grist for discussion and reinterpretation of information, consistent with Wiersema and Bantel’s (1992) finding that heterogeneity is associated with change. This reinterpretation may lead to improved accuracy of subsequent managerial perceptions. The reinterpretation process, however, may be costly in terms of TMT time and team social dynamics. The cost may be worth the investment when the new venture TMT begins with inaccurate perceptions but for TMTs that begin with accurate perceptions, the marginal benefits of reinterpretation may not be substantial enough to offset any potential costs because these teams do not require as much reinterpretation of perceptions. Collectively, the discussion above suggests the following proposition:
P3. TMT functional heterogeneity will moderate the relationship between inaccuracy of initial perceptions and venture performance: TMT functional heterogeneity will have a more positive effect on the venture performance of entrepreneurs who had inaccurate initial perceptions than on the venture performance of entrepreneurs who had more accurate initial perceptions.
Organizational Structure Factors
In addition to TMT characteristics, organizational structure may dictate the extent of double loop learning that a firm-level information processing system can achieve (Daft & Weick, 1984; Rajagopalan, Rasheed, & Datta, 1993). Some researchers have studied the relationship between organiza- tional characteristics and venture performance (e.g., Covin & Slevin, 1990).
MARK SIMON ET AL.
146
In the following section, we propose that two structure factors – decentralization and organicity – may affect the information processing systems of new ventures.
Decentralization
Decentralization refers to the extent to which major decisions are made at a lower versus higher level in the organization (Van de Ven, Hudson, &
Schroder, 1984). Most of the entrepreneurship studies examining decentra- lization found that it improves the performance of new ventures (Van de Ven et al., 1984). For example, Duchesneau and Gartner (1990) determined that successful entrepreneurs encouraged participative decision making at the strategic and operational levels and shared command with lower ranking employees.
We further argue that decentralization is especially important for entrepreneurs who relied on a heuristics-based decision logic to initiate the venture because decentralization may facilitate a double loop learning information processing system within the firm (Milliken, 1990). Sutcliffe (1994), for instance, found that decentralization improves the accuracy of a firm’s perceptions of its environment. This may happen because of broader noticing available to the firm, where multiple and diverse boundary spanners are participating in decision making. Because decision making is decentralized, individuals who reinterpret causal relationships may have more autonomy to initiate new actions based on this changed perception.
Greater centralization, in contrast, narrows the firm’s perspective to that of the entrepreneur and his or her initial misperceptions, which may increase commitment to past actions (Staw, 1981) rather than encourage new actions. Finally, in decentralized firms, because more people are responsible for noticing, feedback from action may be more immediately apparent, increasing the cycling speed of the information processing system. Thus, decentralization accomplishes two goals: first, it may increase the firm’s ability to notice and accurately interpret relevant feedback, and second it may grant others in the organization the power to refine strategic initiatives based upon that feedback, thereby increasing the firm’s competitive advantage. Proposition 4 follows:
P4. Decentralization will moderate the relationship between inaccuracy of initial perceptions and venture performance: Decentralization will have a more positive effect on the venture performance of entrepreneurs who had inaccurate initial perceptions than on the venture performance of entrepreneurs who had more accurate initial perceptions.
Organicity
Another organizational factor studied in the new venture literature is the extent to which a firm is organic or mechanistic (Burns & Stalker, 1961;
Covin & Slevin, 1990). An organic firm is characterized by informality, open communication channels, and adaptiveness. Mechanistic firms, in contrast, have formal structures, restricted communication channels, and consistent actions. Interestingly, entrepreneurship studies have reached contradictory results regarding the effects of an organic structure on venture performance, with different justifications for their findings, suggesting there may not be a direct relationship between venture performance and level of organicity. For example, Stuart and Abetti (1987) found that organic structures decreased performance. They explained that control was more important than the ability to adapt because new venture activities were very chaotic.
In contrast, after finding that successful firms were more organic than less successful firms, Duchesneau and Gartner (1990) argued that an organic structure helped the firm cope with changing environmental conditions.
Although venture performance is critically important for young firms, effective learning is another outcome that facilitates long-term success. High organicity, we believe, will facilitate double loop learning. The explanations above related to performance suggest that an organic structure becomes especially important when entrepreneurs base initial actions on mispercep- tions because organic structures emphasize learning. If one starts with accurate perceptions, the control provided by a more formal structure will help the firm stay on a predetermined path. If, however, perceptions were inaccurate, an organic structure facilitates needed communication to enhance reinterpretation and allows the firm to make the changes needed to adapt to unfolding truths. That is, organicity is especially important for both reinterpretation and new action. This assertion is consistent with Dougherty’s (1990) finding that firms that introduce new products to new markets succeed by refining their perceptions through multiple cycles in which individuals in different functional areas and at different organiza- tional levels share their understanding. Proposition 5 follows:
P5. Organicity will moderate the relationship between inaccuracy of initial perceptions and venture performance: Organicity will have a more positive effect on the venture performance of entrepreneurs who had inaccurate initial perceptions than on the venture performance of entrepreneurs who had more accurate initial perceptions.
Returning briefly to our example from business practice, we believe SHN illustrates some of the proposed relationships. In the case of SHN, insights MARK SIMON ET AL.
148
from a heterogeneous group including the TMT, as well as Board members who were consulted, helped SHN resolve its crisis. Once a solution began to emerge, a 25-year-old veteran of the industry, where SHN had shifted its attention, was hired for his experience and contacts. The decentralized and organic structure allowed the venture to act on this change in a timely fashion.
DISCUSSION
The essence of entrepreneurship is perceiving and acting upon opportunities (Bygrave & Hofer, 1991; Venkataraman, 1997). More specifically, some suggest that innovative learning, alertness, or heuristic-based cognitions allow entrepreneurs to perceive opportunities in discontinuous environ- mental change that others do not perceive (e.g., Cope, 2003, 2005). The performance ramifications of these actions, however, may depend upon whether one has accurate or inaccurate perceptions about the nature of the opportunity. The very novelty of the opportunity suggests that entrepre- neur’s initial perceptions may, in part, be erroneous. Entrepreneurial insights that are difficult to analyze or are based on highly unusual new combinations do not lend themselves to a conventional ‘‘planning and analysis’’ approach. Thus, paradoxically, ‘‘misperceived’’ start-ups may be the rule rather than an exception in innovative ventures.
Recently, however, a study of major leading companies indicated that accuracy of perceptions at start-up, as reflected in the quality of initial product/service offerings, may not be critical for long-term success (Collins &
Porras, 1997). This suggests that founders must draw on other resources to be successful. How, then, do start-up entrepreneurs nevertheless succeed? We suggest that it is a function of radical, double loop learning across several stages and levels of analysis in the process. In the earliest stages of venture formation, an entrepreneur’s capacity for innovative learning is critical: those that can see an opportunity that others do not are more likely to potentially gain competitive advantage. Once the firm grows beyond the founders, however, firms with information processing systems that facilitate double loop learning are needed to correct any early misperception. This latter area is the focus of this paper. We have identified four aspects of a double loop learning information processing system that we believe provide a strategic approach to managing misperceived start-ups.
Specifically, we offered propositions suggesting that a more experienced and diverse TMT adds breadth and depth to a firm’s learning processes that